| 61 | 0 | 888 |
| 下载次数 | 被引频次 | 阅读次数 |
当前电力通信网络规模庞大、设备关联性强，随着网络扩展和业务复杂化，保障其稳定高效运行成为关键挑战。现有方法难以充分捕捉不同业务场景下的路径时空相关性，限制了风险预测与传输效率评估。为此，本文提出一种基于时空拓扑关联的风险评估方法，结合双向长短期记忆网络（bidirectional long short-term memory,Bi-LSTM）和图卷积神经网络。首先，使用Bi-LSTM自编码器提取多变量时间序列的时序特征；其次，结合通信拓扑与先验知识构建电力通信知识图谱；然后，在双曲空间中应用卷积聚合提取通信实体的结构特征；随后，通过注意力机制融合时序与结构特征，为物理节点生成风险评分，并完成不同服务路径的风险评估。结果表明，该方法在AUC(area under the curve)、ACC(accuracy)和F1分数等指标上优于主流图神经网络和时序预测方法，提升了电力通信网络的风险评估能力，可有效降低通信链路或设备故障对数据传输的影响，保障电网稳定运行。
Abstract：The current power communication network is large in scale and has strong equipment associations. As the network expands and business complexity increases, ensuring stable and efficient operation has become a critical challenge. Existing methods struggle to fully capture the spatiotemporal correlations of paths in different business scenarios,limiting risk prediction and transmission efficiency evaluation. To address this,we propose a risk assessment method based on spatiotemporal topological correlation,integrating bidirectional long short-term memory(Bi-LSTM) and graph convolutional networks. First,a Bi-LSTM autoencoder is used to extract temporal features from multivariate time series data. Next,a power communication knowledge graph is constructed by combining communication topology with prior knowledge. Then, convolutional aggregation is applied in hyperbolic space to extract the structural features of communication entities. Subsequently, temporal and structural features are fused through an attention mechanism to generate risk scores for physical nodes,completing risk assessments for different service paths. The results show that this method outperforms mainstream graph neural networks and temporal prediction methods in AUC(area under the curve),ACC(accuracy), and F1-score, enhancing the risk assessment capability of power communication networks, effectively reducing the impact of communication link or equipment failures on data transmission,and ensuring the stable operation of the power grid.
[1]刘钰，熊兰，肖丹，等.基于业务重要度的电力通信路由系统可靠性分析[J].电测与仪表，2017,54(12):34-41.
[2]马泽君，陈志杰，陆军，等.电力通信网络安全可靠性定量评估研究方法[J].电力信息与通信技术，2016,14(4):102-107.
[3] Liu S M,Chen H H,Hu Z Y.Research on risk assessment and early warning of supply chain based on extension in the context for new retailing[M]//Proceedings of the Sixth International Forum on Decision Sciences.Singapore:Springer Singapore,2019:1-14.
[4] Abderraouf B,Mohammed B,Mohamed Salah M,et al.Quantitative risk assessment approach to a dynamic safety evaluation:Skikda's coastal city liquefied gas plant[J].International Journal of Performability Engineering,2023,19(1):10.
[5] Chen X J,Xu L L,Li X,et al.Risk assessment method of power communication network based on gray cloud theory and matter-element extension model coupling[J].IEEE Access,2023,12:47581-47593.
[6] Hong S P.Generalized autoregressive score asymmetric laplace distribution and extreme downward risk prediction[J].arXiv preprint arXiv:2008.01277,2020.
[7] Audibert J ,Michiardi P ,Frédéric Guyard,et al.USAD:unsupervised anomaly detection on multivariate time series[C]//KDD'20:The 26th ACM SIGKDD Conference on Knowledge Discovery and Data Mining.ACM,2020.
[8] Wang M,Wang H,Zhang F. FAMC-net:frequency domain parity correction attention and multi-scale dilated convolution for time series forecasting[C]//Proceedings of the 32nd ACM International Conference on Information and Knowledge Management,Birmingham United Kingdom,2023:2554-2563.
[9] Fulcher B D,Jones N S. Highly comparative featurebased time-series classification[J]. IEEE Transactions on Knowledge and Data Engineering,2014,26(12):3026-3037.
[10] Zhang Y ,Karve P M ,Mahadevan S.Power grid operational risk assessment using graph neural network surrogates[J].arXiv preprint arXiv:2311.03661,2023.
[11] Wang Y,Ye F,Li B Q,et al.UrbanFloodKG:an urban flood knowledge graph system for risk assessment[C]//Proceedings of the 32nd ACM International Conference on Information and Knowledge Management,Birmingham United Kingdom,2023:2574-2584.
[12] Xiong C Y,Power R,Callan J.Explicit semantic ranking for academic search via knowledge graph embedding[C]//Proceedings of the 26th International Conference on World Wide Web,Perth Australia,2017:1271-1279.
[13] Dong L,Wei F,Zhou M,et al.Question answering over freebase with multi-column convolutional neuraln etworks[C]//Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the7th International Joint Conferenceon Natural Language Processing(Volume 1:Long Papers). Beijing,China,2015:260-269.
[14] Wang H W,Zhang F Z,Wang J L,et al.RippleNet:propagating user preferences on the knowledge graph for recommender systems[J]. arXiv preprint arXiv:1803.03467,2018.
[15] Cannon J W,Floyd W J,Kenyon R,et al.Hyperbolic geometry[J].Flavors of Geometry,1997,31:59-115.
[16] Linial N,London E,Rabinovich Y. The geometry of graphs and some of its algorithmic applications[J].Combinatorica,1995,15(2):215-245.
[17] Nickel,M,Kiela,D.Poincaréembeddings for learning hierarchical representations[J]. Neural Inf Processing Syst,2017,30:6341-6350.
[18] Haghighi V,Soltani B,Mahmood A,et al. GCN-based multi-task representation learning for anomaly detection in attributed networks[J]. arXiv preprint arXiv:2207.03688,2022.
[19] Li Y,Zhou Z B,Deng S G,et al.Accurate anomaly detection leveraging knowledge-enhanced GAT[C]//2024IEEE International Conference on Web Services(ICWS),Shenzhen,China,2024:568-577.
[20] Pan S,Hu R,Long G,et al. Adversarially regularized graph autoencoder for graph embedding[J]. arXiv preprint arXiv:1802.04407,2018.
[21] Wang X ,He X ,Cao Y ,et al.KGAT:Knowledge graph attention network for recommendation[J]. ACM,2019:3330989.
[22] Ganea O,Bécigneul G,Hofmann T. Hyperbolic neural networks[J].Advances in Neural Information Processing Systems,2018,31.
[23] Kipf T N ,Welling M. Semi-supervised classification with graph convolutional networks[J]. arXiv preprint arXiv:1609.02907,2016.
[24] Veli V C Kovi C P,Cucurull G,Casanova A,et al.Graph attention networks[J].arXiv preprint arXiv:1710.10903,2017.
[25] Hamilton W L,Ying R,Leskovec J.Inductive representation learning on large graphs[J].Advances in Neural Information Processing Systems,2017:02216.
[26] Homayouni H,Ghosh S,Ray I,et al.An autocorrelationbased LSTM-autoencoder for anomaly detection on time-series data[C]//2020 IEEE International Conference on Big Data(Big Data),Atlanta,GA,USA,2020:5068-5077.
[27] Bian J H,Wang L,Scherer R,et al.Abnormal detection of electricity consumption of user based on particle swarm optimization and long short term memory with the attention mechanism[J].IEEE Access,2021,9:47252-47265.
基本信息:
DOI：10.13774/j.cnki.kjtb.2025.08.012
中图分类号:TM73
引用信息:
[1]孙超,刘磊,田安琪,等.基于时空拓扑关联的电力通信网络风险评估[J].科技通报,2025,41(08):77-86.DOI:10.13774/j.cnki.kjtb.2025.08.012.
基金信息:
国网山东省电力公司科技项目(520627240007)